You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I tried with several pretrained models, including: google-bert-base-en, google-albert-base-en, pai-albert-base-en, they all have the same kind of error:
Key bert_pre_trained_model/bert/encoder/layer_10/attention/output/LayerNorm/beta not found in checkpoint
[[node save/RestoreV2 (defined at /home/jinzy/.conda/envs/et/lib/python3.6/site-packages/easytransfer/engines/model.py:658) = RestoreV2[dtypes=[DT_FLOAT, DT_FLOAT, DT_FLOAT, DT_FLOAT, DT_FLOAT, ..., DT_FLOAT, DT_FLOAT, DT_FLOAT, DT_FLOAT, DT_INT64], _device="/job:localhost/replica:0/task:0/device:CPU:0"](_arg_save/Const_0_0, save/RestoreV2/tensor_names, save/RestoreV2/shape_and_slices)]]
The text was updated successfully, but these errors were encountered:
After empty the 'checkpointDir', it can train without errors. It seems like that every time I changed the pretrained model, the checkpointDir should be emptied, otherwise it will report this error.
I tried with several pretrained models, including: google-bert-base-en, google-albert-base-en, pai-albert-base-en, they all have the same kind of error:
The text was updated successfully, but these errors were encountered: